Merge pull request #15 from thomwolf/master

Update to spaCy 2

Author: thomwolf

neuralcoref/data.py

@@ -527,16 +527,7 @@ def add_utterances(self, utterances, utterances_speaker=None, speakers_names=Non
             utterances_speaker = ((i + a + 1) % 2 for i in range(len(utterances)))
         utterances_index = []
         utt_start = len(self.utterances)
-        for utt_index, (utterance, speaker_id) in enumerate(zip_longest(utterances, utterances_speaker)):
-            if utterance is None:
-                break
-            # Pipe currently broken in spacy 2 alpha
-            # Also, spacy 2 currently throws an exception on empty strings
-            try:
-                doc = self.nlp(utterance)
-            except IndexError:
-                doc = self.nlp(u" ")
-                if self.debug: print("Empty string")
+        for utt_index, (doc, speaker_id) in enumerate(zip_longest(self.nlp.pipe(utterances), utterances_speaker)):
             if speaker_id not in self.speakers:
                 speaker_name = speakers_names.get(speaker_id, None) if speakers_names else None
                 self.speakers[speaker_id] = Speaker(speaker_id, speaker_name)

readme.md

@@ -8,28 +8,21 @@ This coreference resolution module is based on the super fast [spaCy](https://sp
 Be sure to check out [our medium post](https://medium.com/huggingface/state-of-the-art-neural-coreference-resolution-for-chatbots-3302365dcf30) in which we talk more about neuralcoref and coreference resolution.
 
 ## Installation
-Clone the repo (the trained model weights are too large for PyPI)
+Clone the repo and install using pip (the trained model weights are too large for PyPI)
 
 ```
+git clone [email protected]:huggingface/neuralcoref.git
 cd neuralcoref
-pip install -r requirements.txt
+pip install .
 ```
 
 
-You will also need an English model for spaCy if you don't already have spaCy installed.
+You will also need an English model for spaCy if you don't already have spaCy installed in your environment.
 ````
 python -m spacy download 'en'
 ````
 
-The mention extraction module is strongly influenced by the quality of the parsing so we recommend selecting a model with a higher accuray than usual. Since the coreference algorithm don't make use of spaCy's word vectors, a medium sized spaCy model like  ['en_depent_web_md'](https://github.com/explosion/spacy-models/releases/en_depent_web_md-1.2.1) can strike a good balance between memory footprint and parsing accuracy.
-
-To download and install the  ['en_depent_web_md'](https://github.com/explosion/spacy-models/releases/en_depent_web_md-1.2.1) model:
-````
-python -m spacy download 'en_depent_web_md'
-python -m spacy link en_depent_web_md en
-````
-
-If you are an early user of [spacy 2 alpha](https://github.com/explosion/spaCy/releases/tag/v2.0.0-alpha), you can use `neuralcoref` with spacy 2 without any specific modification.
+The mention extraction module is strongly influenced by the quality of the parsing so we recommend selecting a model with a higher accuray than usual.
 
 ## Usage
 ### As a standalone server

requirements.txt

@@ -16,7 +16,7 @@ python-mimeparse==1.6.0
 regex==2017.4.5
 requests==2.18.1
 six==1.10.0
-spacy==2.0.4
+spacy==2.0.7
 termcolor==1.1.0
 thinc==6.5.2
 toolz==0.8.2

setup.py

@@ -17,10 +17,14 @@
         'Programming Language :: Python :: 3.4',
         'Programming Language :: Python :: 3.5'
       ],
-      packages=find_packages(),
-      include_package_data=True,
-      package_data={'neuralcoref': ['neuralcoref/weights/*.npy']},
-
+      install_requires=[
+      'numpy',
+      'spacy',
+      'falcon'],
+      packages=['neuralcoref'],
+      package_dir={'neuralcoref': 'neuralcoref'},
+      package_data={'neuralcoref': ['weights/*.npy',
+                                    'weights/*.txt']},
       keywords='NLP chatbots coreference resolution',
       license='MIT',
       zip_safe=False,